We are only at the beginning of 2025 and the CEO of the technology company Nvidia, Jensen Huang, has already stated that the“the era of agentic AI has arrived”. TheGoogleandOpenAIrecently announced new agency models planned for launch in 2025, and40%of large Brazilian companies plan to integrate agentic AI systems into their operations this year.
How is agentic AI different from its predecessors, and what could the tech buzzword of 2025 mean for Brazilian consumers? The analysts are expecting radical changes, but the final results may be more subtle than expected.
"Artificial intelligence is an extremely exciting new stage in the development of intelligent systems," comments Krishna Tammana, Chief Technology Officer at Gupshup. "But, as with all advanced technologies, it's easy to get caught up in technical jargon." Instead, make a comparison with the size and scale of the shift from the phone to the smartphone in the late 2000s. "If what we are using now, generative AI, is a single application, then agentic AI is like using a smartphone," says Krishna.
Generative AI, like the older cell phone models, is basically a single-use application. You open it, interact with it, and close it until you need it again, the executive continues. If we compare it to a smartphone, which can contain many apps at the same time, that's what agentic AI looks like in relation to generative AI. It's no longer about a single app, but an entire smartphone full of various apps, all interacting with each other at the same time.
The AI agent was also created to perform tasks autonomously, make decisions, and adapt to user information and environmental changes. And, although generative AI has seen use cases for consumers and businesses, this next evolutionary step will likely impact professional life more than personal.
Soon, we will have a workday where the most time-consuming and repetitive tasks are performed autonomously, freeing up more time for us to focus on higher-level strategic, creative, and social functions. In our personal lives, agentic AI can also materialize as a helpful assistant that can place orders and reservations, organize a weekend social schedule, or even manage our finances, for example. However, since no one knows how close we are to large-scale artificial general intelligence, experts are looking for pioneering sectors to lead the way.
"The first sectors to implement agential AI do so behind the scenes in their operations," explains Tammana. "The financial sector is seeing results in fraud detection, doctors and pharmacists focusing on data analysis, research, and development, and manufacturers are creating intelligent robots, and even entire factories, that can operate with minimal human intervention," he observes.
Unless they work in these specialized fields, the daily life of Brazilians may not feel an immediate impact from agentic AI, except in one key area. "At Gupshup, the AI agent handles a range of requests, but one of the most advanced is managing customer interactions," Krishna continues. Therefore, I wouldn't be surprised if, this year, consumers start to notice a difference when talking to customer support agents.
These models are trained with company-specific data, with integrated protection barriers and human agents available for transfer at any time. Therefore, Gupshup's chief technology officer tells consumers to watch out for the shift in interactions between chatbots and humans for a smoother service. "The fact that you don't realize who is human and who is AI means we did a great job. But if the agent you're talking to is experienced, helpful, proactive, and available 24/7, then this could be your first interaction with an AI agent," he explains.
In Brazil, a partnership with a major national fashion brand highlights the emerging landscape of agentic AI. Gupshup is among the first conversational AI platforms to launch an AI agent in Brazil for the retailer Reserva. The brand's AI agent offers users a unique shopping experience, with AI available directly on WhatsApp. Users can search for products by sending text or voice messages, and even describe the characteristics of a person they want to gift. This was done on Mother's Day, when the user had the chance to detail what their mother was like and thus find the best gift to celebrate the occasion.
In this way, the AI understands and recommends the products that best fit the description provided by the user, who can easily check the available products through a native WhatsApp carousel model, having access to more information about the product, or even being redirected to the website to complete the purchase.
